Research map / Detection and segmentation / Open-set and few-shot detection
The Mapillary Traffic Sign Dataset for Detection and Classification on a Global Scale
Christian Ertler, Jerneja Mislej, Tobias Ollmann, Lorenzo Porzi, Gerhard Neuhold, Yubin Kuang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| pre-train on MTSD instead of ImageNet init (ablation) | ResNet50 FPN detector · TT100K detection (max 2048px) | AP: 91.27 -> 97.60 (+6.33) | Improved |
| pre-train on MTSD instead of ImageNet init (ablation) | ResNet50 FPN detector · MVD traffic-sign detection (max 4000px) | AP: 72.90 -> 76.31 (+3.41) | Improved |
| pre-train on MTSD instead of ImageNet init (ablation) | ResNet50 FPN detector · MVD traffic-sign detection (max 2048px) | AP: 64.00 -> 68.29 (+4.29) | Improved |
| pre-train on MTSD instead of ImageNet init (ablation) | ResNet50 FPN detector · TT100K detection (max 2048px) | AP_small: 84.01 -> 93.13 | Improved |
| pre-train on MTSD instead of ImageNet init (ablation) | ResNet50 FPN detector · MVD traffic-sign detection (max 2048px) | AP_small: 30.70 -> 33.60 | Improved |
| use ResNet101 instead of ResNet50 backbone (ablation) | FPN detector · MTSD detection (max 4000px) | AP: 87.3 -> 88.44 | Improved |
| use ResNet101 instead of ResNet50 backbone (ablation) | FPN detector · MTSD detection (max 2048px) | AP_small: 52.31 -> 56.55 | Improved |
| increase inference resolution from 2048px to 4000px (ablation) | ResNet50 FPN detector · MTSD detection | AP_small: 52.31 -> 73.03 | Improved |
| use ResNet101 instead of ResNet50 backbone (ablation) | FPN50/101 + classifier · MTSD classification (313 classes) | mAP: 81.1 -> 83.4 | Improved |
| decouple a shallow classifier from the binary detector instead of multi-class head | two-stage baseline · TT100K detection+classification | mAP: 81.6 -> 89.9 (+8.3) | Improved |
| add MTSD pre-trained detector (ablation) | FPN50 + classifier · TT100K detection+classification | mAP: 89.9 -> 93.4 | Improved |
| add MTSD pre-trained classifier on top of detector (ablation) | FPN50 + classifier · TT100K detection+classification | mAP: 93.4 -> 95.7 | Improved |
Closest papers by meaning
- Traffic-Sign Detection and Classification in the Wild (CVPR 2016 · similarity 0.92)
- Driving by the Rules: A Benchmark for Integrating Traffic Sign Regulations into Vectorized HD Map (CVPR 2025 · similarity 0.87)
- The Mapillary Vistas Dataset for Semantic Understanding of Street Scenes (ICCV 2017 · similarity 0.87)
- Mapillary Street-Level Sequences: A Dataset for Lifelong Place Recognition (CVPR 2020 · similarity 0.86)
- Traffic Scene Parsing through the TSP6K Dataset (CVPR 2024 · similarity 0.84)
- Sketch and Refine: Towards Fast and Accurate Lane Detection (AAAI 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
